For an institution like the U.S. Supreme Court to

On the Ideological Foundations of Supreme Court Legitimacy in the American Public Brandon L. Bartels Christopher D. Johnston George Washington University Duke University Conventional wisdom says that individuals ideological preferences do not influence Supreme Court legitimacy orientations. Most work is based on the assumption that the contemporary Court is objectively conservative in its policymaking, meaning that ideological disagreement should come from liberals and agreement from conservatives. Our nuanced look at the Court s policymaking suggests rational bases for perceiving the Court s contemporary policymaking as conservative, moderate, and even liberal. We argue that subjective ideological disagreement incongruence between one s ideological preferences and one s perception of the Court s ideological tenor must be accounted for when explaining legitimacy. Analysis of a national survey shows that subjective ideological disagreement exhibits a potent, deleterious impact on legitimacy. Ideology exhibits sensible connections to legitimacy depending on how people perceive the Court s ideological tenor. Results from a survey experiment support our posited mechanism. Our work has implications for the public s view of the Court as a political institution. For an institution like the U.S. Supreme Court to render rulings that carry authoritative force, it must maintain a sufficient reservoir of institutional legitimacy, or diffuse support, with the American public and the other branches of government. 1 Legitimacy is conferred upon Congress and the presidency vis-à-vis an explicit electoral connection and enumerated constitutional powers. On the other hand, the Supreme Court s justices are unelected, they serve life terms and are virtually free from public accountability, they possess no explicit mechanisms for enforcing their rulings, and the Court s chief power judicial review is not derived from the Constitution but instead from the Court s own and most enduring precedent, Marbury v. Madison. Because of these realities, scholars and commentators agree that it is crucial for the Court to maintain a reputation from the public as impartial, trustworthy, and above the politics and bargaining characterizing Congress and the presidency. These ingredients of legitimacy, which help facilitate compliance and reduce demands for fundamental institutional changes, are especially important in the face of policy disagreement with the Court s decisions (Gibson, Caldeira, and Spence 2003b, 2005). Empirically, the Court tends to enjoy robust levels of legitimacy from the public (e.g., Gibson 2007). Because institutional legitimacy is so vital to the Court, it is important to understand the foundations on which Americans ascribe legitimacy to the Court. A crucial question concerns whether legitimacy is a function of individuals ideological preferences in relation to the Court s outputs. A conventional wisdom has emerged that no such relationship exists (e.g., Caldeira and Gibson 1992; Gibson 2007; Gibson, Caldeira, and Spence 2003b), and legitimacy is instead rooted within certain democratic values and awareness of and exposure to the Court vis-à-vis positivity bias (Caldeira and Gibson 1992; Gibson 2007; Gibson and Caldeira 2009a, 2009b; see also Caldeira and Gibson 1992; Casey 1974; Gibson, Caldeira, and Baird 1998; Gibson, Caldeira, and Spence 2003b). Some cracks have begun to emerge in this conventional wisdom, albeit in a somewhat counterintuitive Brandon L. Bartels is Assistant Professor of Political Science, George Washington University, 2115 G Street NW, Suite 440, Washington, DC 20052 (bartels@gwu.edu). Christopher D. Johnston is Assistant Professor of Political Science, Duke University, Campus Box 90204, Durham, NC 27708 (christopher.johnston@duke.edu). We thank James L. Gibson, Eric D. Lawrence, John Sides, anonymous reviewers, and participants at the MIT American Politics Workshop and the George Washington University Junior Faculty Colloquium for helpful feedback and suggestions. The Annenberg Supreme Court Study Data can be obtained from www.annenbergpublicpolicycenter.org. Data for the survey experiment examined in this article were collected through NSF Grant 0094964. Data analyzed in the article are available at http://home.gwu.edu/ bartels/ajps legitimacy.html. 1 Caldeira and Gibson (1992) conceptualize legitimacy as synonymous with diffuse support (Easton 1965). American Journal of Political Science, Vol. 57, No. 1, January 2013, Pp. 184 199 C 2012, Midwest Political Science Association DOI: 10.1111/j.1540-5907.2012.00616.x 184

IDEOLOGICAL FOUNDATIONS OF SUPREME COURT LEGITIMACY 185 manner. Hetherington and Smith (2007) show that in spite of the Court becoming more conservative since the 1970s, conservatives are actually less supportive of the contemporary Court than liberals. 2 Hetherington and Smith argue that a learning lag exists whereby most of the mass public has not correctly perceived the right turn in the Court s policymaking, and most individuals hold a 1960s, Warren Court-era view of the Court as a liberal protector of civil liberties and rights. This lagged perception persists because the media do not provide extensive coverage of the Court s contemporary policymaking and because of the mass public s generally low information levels. The evidence challenges conventional wisdom and suggests an effect opposite to what would be expected if citizens were rationally mapping their ideological preferences onto the contemporary Court s purportedly conservative tenor. On top of this result, Gibson (2007, 528, fn. 25) finds that highly aware liberals possess greater legitimacy orientations than highly aware conservatives. Egan and Citrin (2009) find that exposure to liberal Court decisions significantly decreases legitimacy among conservatives, while exposure to conservative decisions does not decrease legitimacy among liberals. We contend that work by Gibson and colleagues and Hetherington and Smith has relied on a dubious assumption that the contemporary Court is objectively conservative in its policymaking, meaning that ideological disagreement should come from liberals and, to a lesser degree, moderates. Our nuanced review of the Court s policymaking over time suggests a rational basis for citizens perceiving the contemporary Court as conservative, moderate, and even liberal. Therefore, ideological disagreement with the Court could come from individuals across the ideological spectrum. We introduce the concept of subjective ideological disagreement whether individuals perceive they are in ideological disagreement with the Court and argue that it significantly diminishes Supreme Court legitimacy within the public. Inquiries into the ideological foundations of Supreme Court legitimacy have important theoretical, empirical, and normative implications. That liberals and conservatives may possess significantly different levels of legitimacy implies that individuals grant or deny the Court legitimacy based on the ideological tenor of the Court s policymaking. That possibility is antithetical to the traditional theoretical conception of legitimacy, which suggests that legitimacy is not contingent on what the Court 2 It is important to note that Hetherington and Smith (2007) do not measure institutional legitimacy, per se, but instead a more generic form of support for the Court (see also Gibson, Caldeira, and Spence 2003; Grosskopf and Mondak 1998; Hoekstra 2000). does but instead on the Court s standing as an authoritative institution that renders rulings in a legalistic manner. While academic and media accounts highlight how justices decisions often turn on political, ideological, and strategic considerations (e.g., Epstein and Knight 1998; Liptak 2010; Maltzman, Spriggs, and Wahlbeck 2000; Segal and Spaeth 2002), a myth of legality has been thought to persist within the citizenry (e.g., Baird and Gangl 2006; Casey 1974; Gibson 2007). Recent research suggests that many Americans actually view the Court in political and ideological terms (e.g., Bartels and Johnston 2012; Gibson and Caldeira 2011; Scheb and Lyons 2000). An ideological foundation to the Court s legitimacy further enhances the view of the Court as a political institution. Such a foundation also leaves the Court vulnerable to attacks from the public and politicians in light of unpopular decisions (e.g., Clark 2011). The lack of an ideological foundation, however, suggests that even when people disagree with the Court s decisions, they maintain respect for the authority of the Court and its rulings. Assessing the Ideological Tenor of the Supreme Court The assumption that the contemporary Supreme Court is objectively conservative implies that (1) individuals uniformly perceive the Court as conservative; (2) therefore, if ideology has any rational relation to legitimacy, then conservatives should necessarily be more supportive of the Court than liberals (Gibson 2007); and (3) individualswhomayviewthecontemporarycourtasliberalhold misperceptions. For instance, Hetherington and Smith (2007) contend that the reason liberals are more supportive of the Court than conservatives is not due to the Court s contemporary policymaking but instead its liberal legacy from a bygone era. We reassess this assumption by first considering the Court s objective outputs over time. Figure 1a plots the percentage of the Court s decisions that were decided in the liberal direction from the 1953 2008 terms 3 (Spaeth 2008). 4 The smoothed, solid line is a nonparametric lowess line of best fit. 3 The Court s terms last from October of the term s year until the following September. 4 In civil liberties and civil rights cases, liberal decisions favor individuals claiming liberties or rights over governmental restrictions of those liberties or rights; conservative decisions favor such governmental restrictions. In economics cases, liberal decisions favor degrees of governmental involvement in the economy and private business, while conservative decisions reflect a laissez-faire approach (Spaeth 2008).

186 BRANDON L. BARTELS AND CHRISTOPHER D. JOHNSTON FIGURE 1 Supreme Court Liberalism for All Decisions (A) and Salient Decisions (B), 1953 2008 Terms A Percentage of Liberal Decisions B Percentage of Liberal Decisions 100 90 80 70 60 50 40 30 20 10 0 100 90 80 70 60 50 40 30 20 10 0 Warren Court All Decisions 1953 1955 1957 1959 1961 1963 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 Warren Court Burger Court Term of the Court Salient Decisions Burger Court 1953 1955 1957 1959 1961 1963 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 Term of the Court Rehnquist Court Rehnquist Court Roberts Court Roberts Court Note: Salient cases are those that were covered on the front page of the New York Times the day after the Court s decision (Epstein and Segal 2000). Figure 1a reveals well-documented ideological patterns over time. The Burger Court of the 1970s and 1980s provided a clear right turn to the liberal Warren Court of the 1950s and 1960s, but it still managed to produce moderate to right-of-center policy. The Rehnquist Court was also generally moderate-to-conservative, and thus far the Roberts Court has continued this general pattern. 5 From the Burger era onward, average liberalism hovers around 45%. While it is clear that the Court has taken a right turn since the Warren Court era, the data suggest that the contemporary Court is not as conservative as some might assume, especially given the number of Republican appointees since the Burger Court commenced. 5 As is well known, the Burger and Rehnquist Courts did not significantly roll back the Warren Court revolution as some predicted would happen. Note the number of terms in these eras where the liberal rating hits or surpasses 50%. While we think it is reasonable to assume that a great share of the mass public does not have a systematic grasp of the Court s full range of policy outputs over time (e.g., Hetherington and Smith 2007; Slotnick and Segal 1998), that does not mean that the public does not understand any of what the Court does (e.g., Gibson and Caldeira 2009c). Many of the Court s most important rulings e.g., abortion and other key civil liberties and rights issues receive ample attention from the media and political elites, are important topics in election campaigns, and have facilitated the formation of significant ideological cleavages in American politics. Thus, most relevant to the mass public may be the Court s most salient cases. Figure 1b presents the percentage of liberal decisions from the 1953 2008 terms for salient cases, using Epstein and Segal s (2000) salience measure for whether adecisionwascoveredonthefrontpageofthenew York Times the day after the ruling. 6 What is particularly strikinginfigure1bisthatfor12outofthe19termsduringthe Rehnquist Court era (1986 2004 terms), the Court hit or surpassed a 50% liberal rating in salient cases. 7 During the natural Court era from the 1994 2004 terms, the Court produced more liberal decisions (57%) than conservative decisions in salient cases. During the last eight terms of the Rehnquist Court, the Court s liberal rating was 64%. 8 Section A of the online supporting information discusses some of these high-profile liberal decisions. The Court has also issued its fair share of significant conservative decisions, but one would not necessarily expect a solidly 6 The average number of salient cases per term is about 17. The minimum is 6, and the maximum is 30. The numbers decrease over time, due to the Court s decreasing caseload. The number of salient cases per term from 1994 to 2008 is 10, 16, 11, 7, 12, 21, 18, 10, 11, 9, 12, 11, 6, 13, and 6. We concede that the somewhat small number of salient cases per term presents some limitations to the power of Figure 1b. Yet if people are paying attention to the high-profile decisions, these percentages remain an accurate reflection of the balance of liberal versus conservative decisions. One might also question whether the New York Times overreports liberal decisions because of supposed liberal bias. Regardless of the paper s motivations or biases, this measure is still reflective of how much media and elite discourse there will be for a case, which then filters down to the mass public. Thus, we believe that front-page decisions are still accurate reflections of the most salient decisions. 7 We highlight the contemporary Rehnquist era because our primary survey data come from March and April 2005. 8 Our measure of yearly Court liberalism does not account for agenda change, which means intertemporal comparisons will not be completely accurate (see Baum 1988). However, the measure is appropriate for our study given that the relative share of liberal or conservative decisions is what is ultimately delivered to the public. Media and elites do not necessarily filter out agenda change in their descriptions of the Court as liberal or conservative.

IDEOLOGICAL FOUNDATIONS OF SUPREME COURT LEGITIMACY 187 conservative Court to issue as many significant liberal decisions as it has. Our discussion casts doubt on the assumption that the contemporary Court is objectively conservative, which therefore questions how prior work assesses theimpactofideologicaldisagreementonlegitimacy. The Court s liberal policymaking in salient cases also questions the mechanism underlying Hetherington and Smith s (2007) learning lag story by suggesting a rational basis for perceiving the contemporary Court as liberal. Figure 1b suggests that conservatives and liberals may be rationally incorporating information about the Court s contemporary outputs into judgments of the Court. With an ever-increasing number of Republican appointees being put on the Court (12 of 16 since 1969), and with a current conservative majority on the Court, many conservatives are likely to be disappointed with the Court. 9 We are not arguing that the public uniformly perceives the contemporary Court as liberal but that individuals may have different perceptions of the ideological tenor of the Court s policymaking depending on the decisions or media to which they pay the most attention. Importantly, individuals seemingly possess rational bases for perceiving the Court as conservative, moderate, or liberal. Therefore, ideological disagreement with the contemporary Court may come from individuals who are conservative, moderate, or liberal. Subjective Ideological Disagreement and Legitimacy Building on these arguments, we propose a new approach for assessing ideological disagreement with the Court s policymaking that highlights the importance of subjective perceptions of the Court s ideological tenor. We argue that ideological disagreement should be based on matching an individual s subjective perception of the ideological tenor of the Court s policymaking withhisorherownideological preferences. A conservative individual who perceives that the Court s contemporary policymaking is liberal would be in ideological disagreement with the Court. If this same person perceives that the Court is conservative, she would be in ideological agreement. Put another way, the impact of ideology on legitimacy orientations is conditional on subjective perceptions of the Court s 9 Moreover, Republican-appointed Justices Blackmun, Stevens, and Souter underwent liberal ideological drift. And Justice O Connor received and Justice Kennedy still receives criticism for not being reliably conservative. ideological tenor. There are several advantages to basing ideological disagreement on an individual s subjective perception of the Court s policymaking instead of the researcher s assumption an objective assessment that may or may not be appropriate about the Court s contemporary ideological tenor. First, as we have argued, past work infers that the researcher s objective assessment of the Court s ideological tenor is what citizens perceptions should be if they were correctly assessing the Court s policymaking. We have argued that there are legitimate bases for perceiving the contemporary Court as conservative, moderate, and even liberal. Second, the most important part being captured with the subjective approach is that individuals themselves perceive they are in ideological disagreement with the Court, which we believe is more important to capture than possessing an iron-clad rationale for such perceptions and/or a systematic comprehension of the Court s objective policy outputs. We argue that it is this perceived ideological disagreement that ultimately leads to decreased levels of institutional legitimacy. When citizens perceive that the Court is consistently producing policy that is incongruent with their own ideological preferences, they question crucial aspects of the Court s legitimacy: Is the Court really impartial and legalistic? Can the Court be trusted to make decisions that are best for the country? Should the Court s powers be reduced? Because the Court is not legitimized by formal elections or constitutional enforcement mechanisms, subjective ideological disagreement with the Court s policymaking can have especially deleterious consequences for the Court s legitimacy in the American public. Another advantage of our subjective approach is that whatever conceptions individuals might possess about the terms liberal, moderate,orconservative, they are using those same conceptions to assess both the Court s ideological tenor and their own ideological preferences. In addition, the approach we have advocated is generalizable to any context or time period. Instead of the researcher inferring the ideological tenor of the Court in a given time period, our approach uses an individual s perception of the Court s ideological policymaking in a given context, which means that ideological disagreement is not dependent on the researcher inferring the correct ideological tenor of the Court. Our first hypothesis posits the generalized relationship between subjective ideological disagreement and Supreme Court legitimacy: Hypothesis 1: Individuals whose ideological preferences are incongruent with their subjective perceptions of the ideological tenor of the Supreme Court will ascribe lower levels of

188 BRANDON L. BARTELS AND CHRISTOPHER D. JOHNSTON legitimacy to the Supreme Court than those whose ideological preferences are congruent with subjective perceptions of the Court s ideological tenor. While we expect a potent, across-the-board effect of subjective ideological disagreement, as posited in the first hypothesis, we also expect that the magnitude of that impact will differ across different points of the ideological spectrum.weexpectthatideologicaldisagreementwill have its greatest impact on legitimacy among strong ideologues. Another way to view this interactive dynamic is to assess how the impact of ideology on legitimacy changes depending on whether one perceives the Court as liberal, conservative, or moderate. This is a direct implication ofoursubjectiveapproach.weexpectthatideologywill have predictable effects on legitimacy depending on how individuals perceive the Court s policymaking. Hypothesis 2: For those who perceive the Court as liberal, legitimacy will decrease as one moves from strong liberal to moderate to strong conservative. For those who perceive the Court as conservative, legitimacy will increase as one moves from strong liberal to moderate to strong conservative. For those who perceive the Court as moderate, legitimacy will be the highest among moderate citizens and will decrease as one moves toward each ideological extreme. Before turning to the empirics, we address the possibility that the effects we have hypothesized and find below are due not to the influence of subjective ideological disagreement on legitimacy, but of legitimacy on subjective disagreement, i.e., reverse causality or reciprocal causation. As work in political science attests (e.g., Lupia 1994; Popkin 1991; Sniderman, Brody, and Tetlock 1991), citizens often make judgments with less than full information by relying on heuristics. Brady and Sniderman (1985) suggest that when citizens lack objective information about the characteristics of a political object, they may rely on a combination of their feelings toward that object and their own characteristics to impute the object s qualities. Liked things will be seen as more similar to the self than disliked things. The implication for our study is that citizens may impute the ideology of the Court based on preexisting stores of legitimacy. In the context of the Supreme Court, we know of no theory or empirical evidence concerning whether this type of self-projection occurs. While we think a strong foundation exists for our theory and hypotheses, it is important to address this issue if we wish to be confident in our results. We thus supplement our cross-sectional hypothesis tests with an experimental study where subjective perceptions are manipulated rather than measured, thus removing endogeneity as a concern. We report these results below. The online supporting information, Section C, contains additional analyses that bolster the causal ordering associated with our theory and empirical results. 10 Data, Measurement, and Analysis We first analyze data from the 2005 Annenberg Supreme Court Survey, sponsored by the Annenberg Foundation Trust and directed by the Annenberg Public Policy Center at the University of Pennsylvania and Princeton Survey Research Associates International (PSRAI). Interviews were conducted by Princeton Data Source, LLC, an affiliate of PSRAI. The survey interviewed a national random sample of 1,504 adults between March 17 and April 18, 2005. 11 Question wording and response options for all variables are included in the online supporting information, Section B. Table 1 includes details on variables and measurement. Supreme Court Legitimacy. Our dependent variable is institutional legitimacy, which we treat as synonymous with diffuse support, for the Supreme Court. Gibson and colleagues (Caldeira and Gibson 1992; Gibson, Caldeira, and Spence 2003a) have conceptualized diffuse support as a generalized trust in the Court, a willingness to reject fundamental alterations to the functioning of the institution, and impartiality as opposed to politicization. In line with prior research, we build a multi-item summative legitimacy scale described in Table 1. 12 We note that the 10 Questions also arise about how political awareness figures into this causal issue. In the online supporting information, Section D, we argue and show that awareness of the Court enhances the impact of subjective ideological disagreement on legitimacy orientations; the argument and findings are in line with sophistication interaction work in public opinion research. An alternative, which we also consider in the supporting information, is that political awareness may actually accentuate the projection types of effects that we discuss above. Once again, a lack of theory and empirical evidence regarding this alternative account in the Supreme Court context makes it difficult to assess. As we discuss in the supporting information, we believe that the sophistication interaction story applied to our hypotheses presents the more valid theoretical and empirical perspective. Ultimately, future theoretical and empirical research needs to be done to distinguish these causal processes. 11 RDD telephone interviews were conducted. The response rate was 41%, and the cooperation rate was 50%. Rates were calculated in accord with the American Association for Public Opinion Research (AAPOR) standards. 12 Items 1 4 use an agree-disagree response scale. Item 5 s response scale ranges from a great deal to not at all. Before creating the scale, all items were recoded from 0 to 1, where higher values represent greater legitimacy.

IDEOLOGICAL FOUNDATIONS OF SUPREME COURT LEGITIMACY 189 TABLE 1 Variable and Measurement Information from Annenberg Supreme Court Survey Variables Supreme Court Legitimacy (Dependent Variable) Subjective Ideological Disagreement (Key Independent Variable) Measurement Information Multi-item summative scale of five items: (1) whether it might be better to do away with the Court altogether in the face of Supreme Court rulings with which a majority of the public disagreed; (2) whether the Court gets too mixed up in politics ; (3) whether the Court favors some groups more than others ; (4) whether the Court can usually be trusted to make decisions that are right for the country as a whole ; and (5) how much people trust the Supreme Court to operate in the best interests of the American people. Scale recoded from 0 to 1 (1 = high legitimacy). = 0.71. Perception of Supreme Court s ideological tenor: Judging by its recent decisions, do you think the Supreme Court is generally liberal, generally conservative, or is it making decisions more on a case-by-case basis? Used in conjunction with respondent s ideological identification. (See text for additional measurement details.) Control Variables Party Identification Three-category nominal variable: Republican, Independent, or Democrat (excluded group). Political Trust Respondents were asked how much they trust the federal government as a whole to operate in the best interests of the American people. The variable, which includes four categories, was recoded to range from 0 to 1(1= high trust). Awareness of the Court Multi-item scale based on four objective knowledge questions (whether respondents know who the Chief Justice is, the two justices who were women [O Connor was still on the Court], whether the Court can declare acts of Congress unconstitutional, and if a 5 4 ruling is declarative) and two subjective awareness items (how well respondents understand the Court s rulings and to what extent they follow the Court s decisions). The scale was recoded to range from 0 to 1 (1 = high awareness). = 0.66. Differential Media Exposure How much exposure individuals have to sober (newspapers and network news) versus sensationalist media (political talk radio and cable news). Measure = (talk radio + cable news) (newspaper + network news); recodedtorangefrom0to1(1= high sensationalist). Age Measured in years; recoded 0 1. Race Three-category nominal variable: Hispanic, Black, or White (White is excluded group). Female Female = 1, male = 0. Education Five-point scale based on level of schooling; recoded 0 1 (1 = highly educated) Note: Question wording and response options for all variables are included in the supporting information, Section B. first four items are the same as four of the items recommended by Gibson, Caldeira, and Spence (2003a). The fifth item taps generalized trust, like item four. 13 13 Across studies, Gibson and colleagues have used some different combinations of legitimacy indicators. Subjective Ideological Disagreement. To test Hypotheses 1 and 2, we require measures for both the survey respondent s ideology and the respondent s perception of the ideological tenor of the Supreme Court. To measure respondent ideology, the survey asked how individuals would describe their views in politics today very liberal,

190 BRANDON L. BARTELS AND CHRISTOPHER D. JOHNSTON liberal, moderate, conservative, or very conservative. To measure individuals perceptions of the Supreme Court s ideological tenor, we rely on the measure in Table 1. We take the third response to represent perceptions that the Court takes a moderate course of action in its rulings (a topic to which we will return). Table 2 displays both the cross-tabulation of these two variables (A) and our measurement strategy discussed later (B); both collapse ideology into a three-category variable. 14 First, the totals collapsing across ideological types (in the rightmost column) suggest that there is meaningful variation in perceptions of the ideological tenor of the Court. A majority (53.39%) believes that the Court proceeds in a moderate, case-by-case approach in its rulings. While many legal commentators perceive the Court as conservative in its contemporary policymaking, just over a quarter of the respondents in the survey perceive the Court as such. And interestingly, a sizable share of respondents roughly 20% thinks that the Court is liberal in its rulings. The table also displays how perceptions of the Court s ideological tenor are quite different among liberals, moderates, and conservatives. About 44% of liberals perceive that the Court is conservative and are thus ideologically incongruent with their perception of the Court s policymaking. A bare majority (51%) of liberals believes the Court takes a case-by-case, moderate approach to its decisions, while just 5% perceive the Court as liberal. Roughly 61% of moderate respondents think the Court proceeds in a case-by-case manner. Twenty-eight percent perceive the Court s rulings as conservative, and about 12% as liberal. Among conservatives, the patterns largely mirror those for liberals. Forty percent of conservatives think that the Court has been liberal in its recent rulings and are thus in ideological disagreement with their perception of the Court s policymaking. Nearly 50% of conservatives perceive the Court as moderate, while 11% perceive a conservative Court. These descriptive statistics provide a convincing rationale for examining subjective assessments of the Court s policymaking. Liberals, moderates, and conservatives have distinct perceptions of the ideological tenor of the Supreme Court. 15 The empirical analyses testing Hypotheses 1 and 2 below include more specific strategies for measuring sub- 14 When using the 5-point scale, some of the cell sizes become very small. The 3-point scale helps to see the broader patterns in the data regarding how liberals, conservatives, and moderates perceive the ideological tenor of the Court. 15 We also split the sample into low- and high-awareness groups (based on a median split), and the cross-tabs (paralleling Table 1) for the two groups are quite similar. jective ideological disagreement. We also include relevant control variables (described in Table 1) in the upcoming statistical models, and we note the following. For party identification, the survey did not ask the traditional follow-up question assessing partisan strength and independent leaners. Also, past research argues that evaluations of the Supreme Court are connected to more peripheral factors, such as democratic values and broader governmental orientations (Caldeira and Gibson 1992; Gibson 2007; Hetherington and Smith 2007). While the survey does not include democratic values measures, it does include an indicator of political trust, therefore capturing feelings toward the government writ large. Importantly, we measure awareness of the Supreme Court, which has been shown to enhance legitimacy orientations. We also include a measure of differential media exposure; Johnston and Bartels (2010) have found that exposure to sensationalist relative to sober media depresses legitimacy. We also control for age, race, 16 sex,and education. The Impact of Subjective Ideological Disagreement on Legitimacy (Hypothesis 1) To measure global subjective ideological disagreement capable of testing Hypothesis 1, one option is to simply create a distance measure, i.e., categorize both respondents and their perceptions of Supreme Court ideology as liberal, moderate, or conservative and calculate the distance between the two variables. A question arises regarding how liberals and conservatives treat the caseby-case option when assessing the ideological tenor of the Court s rulings, which has implications for understanding the relationship between subjective ideological disagreement and legitimacy. We assume that moderates are in strong agreement with a Court they perceive as proceeding in a case-by-case manner. But for liberals and conservatives, does a case-by-case approach represent an ideologically moderate course of action falling in between liberal and conservative perceptions thus indicating moderate disagreement with the Court s rulings, as indicated by the simple distance measure? Or does it represent a sort of tacit agreement with the Court s rulings, meaning that liberals and conservatives who view the Court as proceeding on a case-by-case basis would not 16 The baseline category also includes the very small number of respondents who are Asian American (25 people in the mass sample) and those who responded other (51 in the mass sample). Results employing alternative operationalizations of race produced substantively and statistically similar findings.

IDEOLOGICAL FOUNDATIONS OF SUPREME COURT LEGITIMACY 191 TABLE 2 Respondent Ideology by Perceived Supreme Court Ideology: Empirical Cross-Tabulation (A) and Measurement Strategy (B) A. Cross-Tabulation Ideology of Respondent Perceived Ideology of the Supreme Court Liberal Moderate Conservative Total Liberal 18 51 200 269 4.84% 11.51% 40.32% 20.52% Case-by-Case (Moderate) 190 269 241 700 51.08% 60.72% 48.59% 53.39% Conservative 164 123 55 342 44.09% 27.77% 11.09% 26.09% Total 372 443 496 1,311 100% 100% 100% 100% B. Measuring Subjective Ideological Disagreement Ideology of Respondent Perceived Ideology of the Supreme Court Liberal Moderate Conservative Liberal Strong Agreement Moderate Disagreement Strong Disagreement Case-by-Case (Moderate) Tacit Agreement Strong Agreement Tacit Agreement Conservative Strong Disagreement Moderate Disagreement Strong Agreement actually perceive themselves to be in ideological disagreement with the Court? We address these questions below in the process of laying out our measurement strategy depicted in Table 2b. First, coding strong agreement and strong disagreement is straightforward. For the former, one simply identifies correspondences between the ideology of the respondent and the perceived ideology of the Court, which is captured by the white boxes along the diagonal in Table 2b. Strong disagreement is indicated by the black boxes; it includes liberals who perceive the Court as conservative and conservatives who perceive the Court as liberal. Moderates who perceive the Court as liberal or conservative are categorized as in moderate disagreement. Where the 4-point scale in Table 2b departs from a simple 3-point distance measure concerns separating out what we call tacit agreement from moderate disagreement. Note that the 3-point distance scale would categorize the light-gray tacit agreement boxes as moderate disagreement. However, we believe that liberals and conservatives who respond with the case-by-case option are not expressing subjective disagreement with the Court s rulings but are instead indicating a sort of implicit agreement with the Court s rulings, albeit a weaker type of agreement compared to strong agreement. Descriptive evidence is provided in Figure 2, which graphs group means of legitimacy for categories of subjective ideological disagreement using the simple 3-point distance measure (Figure 2a) and the 4-point measure advanced in Table 2b (Figure 2b). Figure 2a shows that, when using the 3-point measure, increases in subjective ideological disagreement are associated with decreasing legitimacy orientations. Figure 2b provides support for our splitting apart the tacit and moderate disagreement categories. There is again strong descriptive evidence that subjective ideological disagreement depresses legitimacy orientations. Moreover, note how the degree of legitimacy is highly similar between the strong agreement and tacit agreement categories, while legitimacy is a great deal lower for those in moderate relative to tacit agreement. This evidence supports our contention that liberals and conservatives treat the case-by-case option as a sort of tacit agreement as opposed to moderate disagreement. It also supports the decision to use the 4-point measure over the 3-point measure. The frequencies for our 4-point measure are strong disagreement (27.8%), moderate disagreement (13.3%), tacit agreement (32.9%), and strong agreement (26.1%). Because we expect subjective ideological disagreement may have a nonlinear impact on Supreme Court legitimacy, we specify this variable

192 BRANDON L. BARTELS AND CHRISTOPHER D. JOHNSTON FIGURE 2 Comparing Measures of Subjective Ideological Disagreement (Group Means of Supreme Court Legitimacy for Each Subjective Ideological Disagreement Category) A Mean Supreme Court Legitimacy Simple Three-Point Ideological Distance Measure 0.65 0.60 0.55 0.50 0.45 0.40 B 0.65 Mean Supreme Court Legitimacy 0.60 0.55 0.50 0.45 0.40 Agreement Moderate Disagreement Subjective Ideological Disagreement Strong Disagreement Four-Point Measure Separating Out Case-by-Case Perceptions for Liberals and Conservatives Strong Agreement Tacit Agreement Moderate Disagreement Subjective Ideological Disagreement Strong Disagreement as a four-category nominal variable. We dummy out the categories and exclude strong agreement (the baseline category). Table 3 includes results from an OLS model testing whether subjective ideological disagreement significantly decreases Supreme Court legitimacy. The model fit is very good, with an adjusted R 2 of 0.31. Importantly, the results show that subjective ideological disagreement exhibits a statistically significant and substantial impact on TABLE 3OLSModeloftheImpactofSubjective Ideological Disagreement on Supreme Court Legitimacy (Hypothesis 1) Variable Coeff. (SE) p ŷ Subjective Ideological Disagreement Strong Disagreement 0.139 (0.014) 0.000 0.461 Moderate 0.088 (0.017) 0.000 0.512 Disagreement Tacit Agreement 0.014 (0.013) 0.267 0.586 Strong Agreement 0.600 (Baseline Category) Republican 0.006 (0.013) 0.640 Independent 0.006 (0.012) 0.627 Political Trust 0.242 (0.019) 0.000 Awareness of the Court 0.227 (0.024) 0.000 Differential Media 0.085 (0.026) 0.001 Exposure Age 0.033 (0.024) 0.178 Hispanic 0.008 (0.022) 0.725 Black 0.064 (0.017) 0.000 Female 0.025 (0.010) 0.014 Education 0.066 (0.016) 0.000 Intercept 0.413 (0.027) 0.000 Note: OLS estimates; N = 1236; F = 43.60, p <.001; Adj. R 2 = 0.31.ŷ represents the predicted value of legitimacy orientations for each subjective ideological disagreement category while holding the remaining variables constant at their mean values. legitimacy orientations among American citizens, ceteris paribus. A joint F-test shows that the three subjective ideological disagreement coefficients are jointly statistically significant (F = 57.76, df = 3, p <.001). Looking at the effects of the dummies, the results show that strong ideological disagreement that is, conservatives and liberals who think the Court is liberal and conservative, respectively significantly and sizably depresses legitimacy orientations, relative to those in ideological agreement. Since the legitimacy scale is coded from 0 to 1, this impact of 0.139 represents just over oneseventh of the scale. The ŷ column in Table 3 lists the predicted values of legitimacy for each subjective ideological disagreement category while holding the remaining variables constant at their mean values. Paralleling the descriptive results from Figure 2b, the results show that when controlling for the other independent variables, individuals with strong subjective disagreement possess moderate to low levels of institutional legitimacy (0.46), while those

IDEOLOGICAL FOUNDATIONS OF SUPREME COURT LEGITIMACY 193 in subjective ideological agreement possess much higher levels of legitimacy (0.60). Next, those in moderate ideological disagreement i.e., moderates who believe the Court is either liberal or conservative also possess significantly lower levels of Supreme Court legitimacy relative to those with strong ideological agreement. The decrease of.088 represents about one-eleventh of the legitimacy scale, suggesting a substantively meaningful effect. Those with moderate disagreement have a predicted legitimacy value of.51, which is significantly greater than the value for individuals with strong ideological disagreement (p <.01). 17 Individuals in moderate ideological disagreement also have significantly lower legitimacy levels than those who are in tacit agreement (p <.001), which again suggests the appropriateness of using the four-category measure. Recall that tacit agreement captures individuals who are liberal or conservative and believe that the Court proceeds on a case-by-case basis. The small and statistically insignificant difference (.014) between the tacit agreement and strong agreement groups suggests that liberals and conservatives who assess the Court as issuing rulings on a case-by-case basis are treating that perception as akin to outright ideological agreement. We briefly discuss results for the control variables. First, after controlling for subjective ideological disagreement (and other factors), party identification exhibits no impact on legitimacy. Political trust exhibits a statistically significant and sizable impact. In line with prior research, awareness exhibits a quite potent impact. In support of Johnston and Bartels (2010), as the balance of media exposure tilts toward sensationalist relative to sober coverage, legitimacy significantly decreases. The results confirm prior work showing that blacks ascribe lower legitimacy to the Court than do whites (Gibson and Caldeira 1992). Moreover, legitimacy is significantly lower for females than for males, though the effect is not substantial. Finally, akin to the effect for awareness, highly educated individuals hold significantly higher degrees of legitimacy than those with lower levels of education. On the whole, the results are supportive of Hypothesis 1: subjective ideological disagreement significantly depresses legitimacy orientations among Americans. While much of the conventional wisdom casts doubt on the role of ideological preferences in shaping legitimacy orientations, our approach of using subjective judgments of ideological disagreement with the Court uncovers a potent ideological foundation to Supreme Court legitimacy in the mass public. 17 This inference was made by changing the baseline category. FIGURE 3 Impactof Ideology on Legitimacy, Conditional on Perceptions of the Supreme Court s Ideological Tenor (Hypothesis 2) Court perceived as liberal Court perceived as taking case-by-case approach Court perceived as conservative Predicted Supreme Court Legitimacy.35.4.45.5.55.6 VL L M C VC Ideology of Respondent Note: VL= Very Liberal, L = Liberal, M = Moderate, C = Conservative, VC = Very Conservative. The Impact of Ideology Conditional on Ideological Perceptions of the Court (Hypothesis 2) WenowtestHypothesis2 howtheimpactofideology on legitimacy changes depending on how one perceives the Court s ideological tenor. Instead of using our measure of subjective ideological disagreement, we interact respondents ideology (using the 5-point ideology measure) with respondents perception of Supreme Court ideology. To accommodate the potential nonlinear impact for those perceiving the Court as moderate, we use a quadratic operationalization of ideology, including both ideology and ideology squared. This operationalization is costless; if the effect is indeed linear, the quadratic specification is capable of reducing to a linear impact. We treat perceptions of the ideological tenor of the Supreme Court as a three-category nominal variable; we include the liberal and conservative dummy variables; case-by-case basis is the baseline category. To test Hypothesis 2, we interact both ideology and ideology squared with the liberal and conservative perception dummies. We include the same controls as in Table 3. Since the substantive interpretations pertaining directly to Hypothesis 2 are not immediately apparent from the raw model results, we report those results in the online supporting information, Section E, and focus here on Figure 3, which graphically displays the substantive thrust of the analysis. The figure presents predicted values of Supreme Court legitimacy as

194 BRANDON L. BARTELS AND CHRISTOPHER D. JOHNSTON a function of ideology separately for the three perceptions of the ideological tenor of the Court (while holding the remaining variables constant at their mean values). Figure 3 presents substantively powerful findings in support of Hypothesis 2. Individuals map their ideologies onto Supreme Court legitimacy judgments in accord with how they perceive the Court s ideological tenor. Among those who think the Court is liberal in its rulings, individuals describing themselves as very liberal and liberal possess the highest degree of legitimacy, and as ideology moves toward the conservative end of the spectrum, legitimacy significantly and substantially decreases. 18 Among those who perceive the Court as conservative, ideology exhibits the opposite effect compared to those perceiving the Court as liberal. Because they perceive the Court as conservative, strong liberals possess the lowest legitimacy; legitimacy significantly increases as one moves toward the conservative end of the spectrum. 19 While strong conservatives who believe the Court is liberal in its rulings possess the lowest level of Supreme Court legitimacy, strong conservatives who perceive of the Court as conservative exhibit among the highest levels of Supreme Court legitimacy of any group. The results could not more clearly support the need to account for citizens subjective perceptions of the ideological tenor of the Court. Among those who see the Court as deciding cases on a caseby-case basis, a nonlinear, ideological strength effect occurs, with legitimacy maximized for moderate individuals and decreasing as ideological strength increases. 20 Moderates who perceive the Court as taking a case-bycase approach to its rulings maintain the highest degree of legitimacy among the groups examined here. Figure 3 clearly demonstrates that once accounting for how individuals perceive the Court s ideological tenor, ideology exhibits sensible and potent effects on legitimacy. Figure 3 also highlights how conservatives tend to punish the Court for ideological incongruence to a greater extent than liberals. This is seen by examining the vertical distances between the Court perceived as liberal and Court perceived as conservative plots as one moves across the ideological spectrum. The vertical 18 Results using the Clarify software in Stata show that the difference in predicted values of legitimacy between very liberal and very conservative respondents is statistically significant at the a =.05 level. 19 Once again, results using Clarify show that the difference between very liberal and very conservative respondents is statistically significant. 20 Results from Clarify show that the difference between moderate and very liberal is statistically significant, while the difference between moderate and very conservative is not. distances are substantial among conservatives, while minimal for liberals, suggesting that ideological disagreement exhibits a significantly larger effect among conservatives. Conservatives who perceive the Court as liberal maintain the lowest legitimacy orientations, while conservatives who perceive the Court as conservative maintain some of the highest degrees of legitimacy. The findings shed light on prior perspectives seeking to explain why conservatives are less supportive of a supposedly conservative Court than liberals. As discussed already with respect to Figure 1, many conservatives have a rational basis for perceiving the Court as liberal in its contemporary policymaking; those conservatives ascribe very low levels of legitimacy to the Court. Pertaining to Hetherington and Smith s findings, we believe that conservatives who think the Court is liberal possess low levels of legitimacy not because of a learning lag, but because they strongly disagree with the Court s contemporary policymaking. Expectations for a more conservative Supreme Court, given the number of Republican appointments made, likely help to exacerbate the disappointment many conservatives possess about the Court s policymaking. But among conservatives who agree with the Court s contemporary policymaking, perhaps taking into account a wider range of the Court s outputs than just the highly salient decisions, a high degree of legitimacy is ascribed to the Court. This gulf in legitimacy orientations between conservatives in strong agreement and strong disagreement with the Court s policymaking is certainly worthy of additional analysis in the future. Figure 3 also highlights how it is in the best interest of the Court to be perceived as taking a case-by-case, moderate approach to decision making. Among those who perceive the Court as such, legitimacy orientations are consistently the highest or among the highest. And under this condition, ideology does not exhibit a terribly strong impact not nearly as strong as it does for those who perceive the Court as liberal or conservative. In fact, the results show that even conservatives and liberals ascribe more or similar levels of legitimacy to the Court when they believe the Court is acting in a moderate, caseby-case manner compared to acting in congruence with their own ideological leanings. Since those who perceive the Court as undertaking a case-by-case approach make up about half of the public, according to the survey, this finding is significant. For the half perceiving the Court to be on a moderate policy course, ideological strength exhibits only a moderate impact. For the half that sees the Court as liberal or conservative, ideology exhibits potent effects in accord with their perceptions of the Court s ideological tenor.